基于超像素的自然图像抠图

Nature image matting based on superpixels

现有的自然图像抠图算法可以分为三类：基于采样、基于传播和基于机器学习。通常为传播算法设计一个有效的像素特征非常困难,也一直是影响算法结果好坏的重要因素。本文探索将超像素作为特征应用在传播算法上效果,并设计了新的函数来衡量两个超像素之间的相似性。实验结果表明本文提出的方法能更有效地区分前背景像素,建立更准确的全局像素关系,并在标准测试集上取得了领先效果。

Existing image matting methods can be included in three main categories： sampling-based, propagation-based, and machine learning-based. While it is difficult to design an effective pixel feature for the propagation-based methods, which has always been an important factor affecting the results of the algorithm. In this paper we investigate the effect of applying superpixels as a feature on the propagation algorithm, and design a new function to measure the similarity between two superpixels. The experimental results show that the method can effectively distinguish the foreground pixels from background pixels, and establish a more accurate global pixel relationship. The final results achieve a leading effect on the standard benchmark datasets.